Apex Insider Blog

After writing The Peak Age for an NFL Running Back, I became intrigued with how age affects running backs and how predictable the drop-off is for them. In addition, I have recently become fascinated by data analysis programming languages, such as Python and R. The intersection of these two interests led to an attempt to create a predictive model using running back age and PPR scoring.

Defining a Peak Season

To win fantasy football championships, we are all attempting to identify quality or peak running back seasons — difference-making players we can rely on to start each week. Over the past two seasons, the RB10 has averaged 230 PPR fantasy points, which amounts to 14.4 fantasy points per game.

For the purpose of this study, I included all running back seasons since 2000 to score at least 230 PPR fantasy points or at least 200 PPR fantasy points and 14.4 fantasy points per game. Then I charted whether the player maintained the same pace (14.4 fantasy points per game) in the following season. The study includes 223 individual running back seasons.

Note: A fair number of seasons were removed because of inconclusive results. For example, Jamaal Charles’ 2011 season was not included because in the following season he was injured in Week 2. If results for the following season were not conclusive, the season was removed.

The table below contains all of the seasons that were included to create the models. The ‘0’ or ‘1’ under N + 1 indicates whether the player succeeded in the following season.

Preface

In my short education in machine learning, I have learned the following: First and foremost, you must build and teach a model with a training set. Then, to analyze the accuracy of the model, some observations must be set aside to test the model.

In the Kernel SVM, I left out 20 percent of the observations to test the model. Therefore, 179 of the 223 seasons were used as a training set. The remaining 20 percent, or 44 observations, were used to test the model.

Based on the type of data being analyzed, the two best algorithms appear to be Kernel SVM and Naïve Bayes. We’ll start with the Kernel SVM.

Kernel SVM Model

Let’s first understand the chart. The green points indicate players who “succeeded” in the following season, the red points indicate those who did not. The Y-axis refers to PPR points per game while the X-axis refers to Age. The top left is young, high-scoring players. The bottom left refers to low-scoring, young players. The top right refers to old, high-scoring players. The bottom right refers to old, low-scoring players.

Note that there are far fewer points on the right side of the graph — 83.8 percent of the seasons took place before age 29. Once you include the age-29 seasons, it accounts for 203 of the 223 seasons — 90.6 percent. It is hard to just make this prestigious list at 28 years old, let alone succeeding in the following age-29 season.

This is the training model, so we will see how the model fares in predicting the remaining 44 observations shortly. The conclusion we can draw so far is the model doesn’t like the older, low-scoring seasons. The drop-off seems to take effect around age 27 for low scoring seasons. By age 29, a running back’s play looks likely to fall off.

Let’s check the results of Kernel SVM Test.

Out of 44 total observations, the Kernel SVM model predicted 31 correctly, which is an accuracy rate of 70.5 percent. While there are some incorrect predictions on the boundary, the model does a decent job at predicting the drop-off.

Naïve Bayes Model

For the Naïve Bayes model, I used 56 observations to test the model or 25 percent of the seasons. 167 observations were used to train the model.

The Naïve Bayes algorithm returns a similar result – as the curve appears to follow to the same line. Let’s check the results of the Naïve Bayes model test.

Out of 56 total observations, the Naïve Bayes model predicted 42 correctly which is an accuracy rate of 75 percent. It was a little more accurate than the Kernel SVM model.

What Does This Mean for 2019?

The model does not like older running backs who are coming off low scoring seasons.

Heading into last season, there were nine players who matched the peak criteria. Only three – Carlos Hyde, Mark Ingram, and LeSean McCoy were over 25 years old and considered unlikely to maintain their pace. All three did not maintain the 14.4 fantasy points per game pace and underperformed according to their average draft position.

In 2018, 13 players scored at least 230 PPR fantasy points or at least 200 PPR fantasy points and 14.4 fantasy points per game. Saquon Barkley, Christian McCaffrey, Todd Gurley, Alvin Kamara, Ezekiel Elliott, James Conner, James White, Melvin Gordon, David Johnson, Joe Mixon, Tarik Cohen, Kareem Hunt, and Phillip Lindsay. Only three players were over 24 years old last season – Melvin Gordon (25), James White (26), and David Johnson (27).

During the 2019 season, David Johnson will turn 28 years old and scored just 15.5 fantasy points per game in 2018. Obviously, Johnson should be in a much better situation to succeed with Kliff Kingsbury coming to town, but the model does not like his outlook this season.

James White is 27 years old and doesn’t receive as many touches as the other players on this list. Sharing touches with Sony Michel and Rex Burkhead will make it tough for White to continue to produce like an RB1. One silver lining is his production with Rob Gronkowski out.

While Melvin Gordon will be 26 this year, averaging 23 PPR fantasy points per game lands him in the green according to both models. As a workhorse on a good offense, Gordon is currently coming off the board in the middle of the first round.